[Edaily Reporter Yun Junghoon ] Amid recent controversy surrounding NVIDIA CEO Jensen Huang, calls are growing within the U.S. investment community to foster domestic open-source artificial intelligence (AI) companies. Analysts note that there is a growing recognition that an open-source ecosystem backed by U.S. capital is necessary to counter the rapid growth of Chinese AI models.

Lim Jeong-hwan, CEO of Motif Technologies, emphasized during a presentation analyzing the global AI market held on the 12th at the JW Marriott Dongdaemun in Seoul, “Recently, voices among global investors have begun to suggest that the U.S. needs proper open-source AI companies to counter China’s rise in AI.”

CEO Lim explained that he sensed a shift in investment sentiment while meeting with representatives from global venture capital (VC) firms and major investment funds as part of his recent efforts to secure investment from NVIDIA. He noted that while the prevailing view in the past was that open-source AI companies lacked business viability, the strategic value of the open-source ecosystem is now being reevaluated as the U.S.-China technology hegemony competition intensifies.

Offensive by

Low-Cost Chinese Models… OpenAI and Anthropic’s Profitability Put to the Test

Improvements in the performance of Chinese open-source models are threatening the revenue models of leading U.S. AI companies such as OpenAI and Anthropic. In particular, analysts note that as AI usage increases in the business-to-business (B2B) market, customers feeling the financial strain are shifting toward relatively cheaper Chinese models.

CEO Lim analyzed the market situation, stating, “It’s not that new users trying AI for the first time are choosing Chinese models,” and added, “Companies that learned how to use AI through OpenAI and Anthropic’s paid services are switching to cheaper Chinese models because they can no longer bear the cost burden as their traffic increases.”

Currently, the software coding sector is a relatively reliable source of revenue in the AI application market. The revenue of major companies, such as Anthropic, also relies heavily on coding services. However, as Chinese companies are rapidly catching up in this market as well, the challenge of securing new revenue streams is growing.

Only Semiconductors and Infrastructure Are Profitable… “We Must Consider AI Revenue Models”

The revenue imbalance within the AI industry is also a problem. While semiconductor companies—including NVIDIA, SamsungElectronics(005930), and SK hynix(000660) —are reaping the benefits of expanded investment in AI infrastructure, the companies actually developing AI models are shouldering the burden of massive computational costs and operating losses.

Power shortages are another variable. Citing data from an external market research firm, CEO Lim predicted that approximately 40% of the graphics processing units (GPUs) that global hyperscalers have contracted to deploy between 2028 and 2030 could face delays in actual operation due to insufficient power infrastructure.

CEO Lim advises that the Korean AI industry should not focus solely on securing its own foundation models or localizing technology. He argues that securing a revenue structure capable of covering the massive development and operational costs is just as important as competing in model development.

CEO Lim stated, “The most critical issue at this point is where to generate the actual revenue needed to offset the massive costs of model development and operation,” adding, “The competitiveness of the AI industry must be assessed comprehensively—not only in terms of model performance but also in terms of capital-raising capabilities, actual profit-and-loss structures, and the global regulatory and export environment.”